ant final work

8/3/2019 Ant Final Work

1/21

Table of Contents

PART I

1.1 Differential services 2

1.1.1 Advantages 2

1.1.2 Disadvantages 3

1.2 MPLS 3

1.2.1 Advantages 4

1.2.2 Disadvantages . 4

1.3 Quality of services ... 4

PART II Laboratory Work:

2.1 Exercise 1 7

2.2 Exercise 2 .. 8

2.3 Exercise 3 ...10

2.4 Exercise 4 10

2.5 Exercise 5i.. 14

2.6 Exercise 5ii . 16

2.7 Exercise 6.......................................................... 17

2.8 Exercise 8i. 18

2.9 Exercise 8(b) ii.. 20

PART III

Conclusion.. 21

i


2/21

PART I

1.1 Differentiated Services

Basically Differentiated services was introduced to provide end-to-end Quality Of Services (QoS).

Differentiated services architecture is an effective technique to solve the traffic issue in IP

networks by simple separation of packets. Packets entering the differential domain are classified

and marked rather than reservation of packets. Various packet traffic are controlled by router by

applying different per-hop behavior (PHB)in diffserv. It is resource allocation mechanism in which

network service provider offers different levels of services to support variety of application. packet

traffic is marked and routed in service level agreement (SLA). It performs the admission control by

specifying traffic for transmitting in particular flow.[1]

Figure:

conceptual model of DiffServ registration-domain-based wireless network architecture [2]

Expedited Forwarding (EF) and Assured Forwarding (AF) are the two forwarding mechanism used

in Diffserv architecture [3]. Assured Forwarding PHB uses multiple random early detection

(MRED) technique to solve the internet traffic. In wireless networks handoff has higher priority than

new connection hence classification is carried out for priority handoff mechanism. [2]

1.1.1 Advantages:

It removes addition burden on the core network of imposing complex routing techniques.(4)

It can be apply without scalability concern.

Behavior of the network is easy to measure as there is no complexity in the network.

2


3/21

1.1.2 Disadvantages

It is difficult with Diffserv to support end to end quality of services.

Since classification is based on priority mechanism, on congestion low priority packets may

get dropped. [2]

It fails to packet losses in congestion when network capacity is low.

It does not have facility for admission control.

1.2 MPLS

MPLS is basically performs packet forwarding mechanism by adding labels at the edge of thenetwork.It represents the convergence of two fundamentally different approaches that is datagram

and virtual circuit in data networking. Basic fundamental idea of MPLS is to separate control plane

from data plane. Data plane consist of forwarding mechanism which carry out simple label

switching operation and control plane is related to network level coordination function.[5][6] The

main task of MPLS is to provide traffic engineering which ultimately reduces traffic on the network

as well as minimizes cost of the network.

Figure: MPLS label switched path

3


4/21

To manage traffic on the network, at the entrance node of an MPLS domain a fix length label isadded to the packet. MPLS forwarding scheme is based on the label information based (LIBs). Forforwarding incoming packet MPLS checks for the label to be use in its LIB to determine outgoinginterface which is based on forwarding equivalence class (FEC). [6] [7]

1.2.1 Advantages of MPLS

MPLS optimizes Performance of operational IP network.[8] MPLS is deemed tunneling technology which support implementation of virtual private

network (VPN) services.

MPLS technique enhances the service integrity. [9]

For increasing network efficiency MPLS offers traffic engineering and different levels of

schemes. [9]

It is simple as there is no need of managing routing table complexity.

If voice and data compatibility added to MPLS it reduces cost compared to ATM or frame

relay.

1.2.2 Disadvantage of MPLS

It does not support all the equipments of existing IP network

It reduces system overhead by IP over ATM.

Layer 2 point to point offers support for higher bandwidth circuits (more than 10 Mbps).

1.3 Quality of services (QoS) :

Quality of services is the ability to provide level of quality to several applications, priority to

subscribers and information. Quality of services is one of the important issue that service provider

takes into account. By providing better Quality of services network service provider should

maintain high level of subscribers satisfaction.

There are many forms of information can be transmitted such as texts, images , audio and video,

these all together can be called as multimedia.

As popularity of the internet and wireless communication is increasing rapidly, the demands for the

more reliable services is major issue to the network service provider. With the more sophisticated

devices like mobile, laptop etc and high speed world wide access the service providers are

experiencing high demand for multimedia services rather than only voice and data transmission.

Other challenges for service provider in providing multimedia services with better Quality of

services in the wireless transmission are availability of bandwidth resources, effects of different

4


5/21

parameters in environment and mobility of user. MPLS with differentiated services and constraint-

based routing is used to implement various QoS capabilities in IP networks. As explain above

MPLS and differentiated services are used to increase system capabilities and reduce traffic

burden on the network.[10]

Basic issue in transmission of multimedia services is availability of bandwidth and size of the files.

Hence audio and video compression needed, which are done by simple compression techniques.

For compression of audio file MPEG audio encoder and Decoder are used. Video is considered assuccession of still pictures MPEG compression module is used to compress video.[11]

Adaptive resource allocation for multimedia:

In Adaptive resource allocation for different types of applications resources are allocated

adaptively based on different classes.

Figure: Adaptive module for QoS management

To solve the above problem QoS is separated at application-level QoS which deals with the quality

of services provided at user end and connection-level QoS which measures connectivity of service

and continuity of call during handoff. On the basis of delay requirement, multimedia services are

divided into real-time and non real-time services. Bandwidth requirement for real-time services isless hence on its request call is established by virtual circuit concept. Best service concept is used

for non real-time services. Best-effort model for IP fails in providing Video streaming application on

internet in terms of packet losses. Fine-Granular-Scalability scheme for video coding was adopted

to overcome packet losses and to gives the adjustment for irregular and variable bandwidth

between two usres. [12] [13] [14]. Cross layer design technique is used for adaptation of dynamic

behavior of the wireless network with respect to continuously changing environment. [15]. Voice

over IP s basically voice calls made over internet protocol. Voice calls are made with the help of

Public Switched telephone network (PSTN) reducing cost of the system and traffic on the IP

network

5


6/21

Conservative and Adaptive Quality of services (CAQoS)

Conservative and Adaptive Quality of services (CAQoS) was proposed to execute QoS

provisioning for multimedia traffic on network. It increases network capability by maximizing use of

total available bandwidth, reducing handoffs and call blocking. CAQoS performs this by executing

early scaling-down of bandwidth for new connection. [16]

Controlled CIWP (Client-Initiated-With-Prefetchig )

One of the most demanding service in broadband internet service is Video on demand (VOD). For

providing VOD services service provider have to multicast video continuously on the dedicated

channel, which require high bandwidth. To provide VOD service proper utilization of bandwidth is

require, the scheme called controlled CIWP was introduce that support large number of users

while providing instantaneous service by controlling frequency of multicast video. [17]

IP video conferencing

The aim of the service providers in IP video conferencing is to provide high performance audio and

video experience to subscribers. H.323 gatekeeper is used to provide quality of services for the

video conferencing. Mainly gatekeeper performs admission control of the QoS. Priories to the

video conferencing traffic are given by using diffserv code point (DSCP) marking. Admission

control with proper synchronization is executed by the RSVP signaling. For providing efficient

video conferencing all the above elements should be considered together with propersynchronization. [18]

6


7/21

Part-II

Exercise1:

This example simulates two nodes and data traffic over link. The packets are being transfer fromnode0 to node1 by the CBR traffic source. Various analysis have been performed by varying

parameters like packet size, link capacity, delay and packet interval etc.

Result:

1a 1b

1c 1d

Figurelinkcapacity delay

Packetsize

Packetinterval

starttime

endtime

1a 1 Mb 10 ms 500 0.005 0.5 4.5

1b 1 Mb 1 ms 500 0.005 0.5 4.5

1c 1 Mb 10 ms 50 0.005 0.5 4.5

1d 10 Mb 10 ms 500 0.005 0.5 4.5

Above shown values are used to get various results. Different figures in NAM animator will givethe overview of the changes occurs by changing parameters.

7


8/21

Delay: delay is the time taken link to transfer one packet from source to destination. Figure 1b

shows when delay is 1ms packet reaches destination before other packet is been release from

source.

Link capacity: it defines the maximum tolerable traffic on the link. It should be large in order to

avoid losses.

Packet size: figure 1c shows when packet size and interval decreased, efficiency of the systemincreases.

Start and end time: it deals with the start and end of the simulation.

Exercise 2:

This example shows the data transfer between four nodes. Here two types of traffic sources CBR

and FTP also UDP and TCP agent are used to transfer traffic over bottleneck link between n2 and

n3. For bottleneck link, the link capacity should be more greater than sum of the bit rates of the

incoming traffic.

Result:

2a 2b 2c

2d 2e 2f

8


9/21

FigureLink

capacity DelayPacket

sizeCBR

packet rateCBRstart

FTPstart

Queuelimit

2a 1.7 Mb 20 ms 1000 1Mb 0.1 1 10

2b 1.7 Mb 20 ms 1000 1Mb 0.1 1 10

2c 5 Mb 20 ms 1000 1Mb 0.1 1 10

2d 1.7 Mb 5 ms 1000 1Mb 0.1 1 50

2e 1.7 Mb 20ms 1000 3Mb 0.1 1 10

2f 1.7 Mb 20ms 5000 1Mb 0.1 1 10

Above shown values are used to get various results. Different figures in NAM animator gives the

overview of the changes occurs by changing parameters.

Here two traffic source CBR ( constant bit rate) and FTP (variable) are used with two traffic agent

TCP and UDP.

TCP transfer packet only after acknowledge whereas UDP keep sending packets without

acknowledge.

Link capacity: it is clear from figure 2b and 2c where link capacity is different. Figure 2c shows no

packet loss when link capacity is increased to 5Mb.

Queue limit: figure 2d shows the condition where queue limit have increased to 50 hence there is

long queue rather than packet loss.

Packet rate: figure 2e shows the transfer of packets with the rate of 3Mb. As the link capacity is

1.7Mb there is sudden packet loss and there is no share capacity for FTP packets.

Exercise 3:

This example gives more realistic network with bottleneck link. Number of links are established oneither side of the bottleneck link. Link established may be FTP or CBR. Start time and end time for

the traffic is random which makes the network more realistic. In this example FTP links are

established and as start and end time for traffic is random it is difficult to predict what traffic would

be there on link o specific time which make the network realistic. Since this traffic is routed through

bottleneck link which has low link capacity packets are dropped at different rate due to variable

traffic.

Result:

9


10/21

3a 3b

Various parameters are change to get different result.

Number of Nodes: As number of nodes increased traffic on a shared link was increase which

result in high packet loss.

Delay: As delay is increased packet losses increased and vice-versa.

Link capacity: When link capacity increased traffic on link decreases which results into low packet

loss.

Traffic Type: When TCP traffic were used even in case of packet loss packets are resent. In case

of UDP loss were increased.

Exercise 4:

In this example we are going to measure performance metric of the network. There are four types

of performance metric discussed here they are throughput, packet loss, delay and jitter.

Throughput: It is very important performance factor that figures out the capability of link to

transmit particular data. Link capacity has major influence over throughput as link capacity was

increased throughput of the link were increased. However link delay did not affect much on the

throughput. When the traffic of UDP was replace by TCP traffic throughput overall performance

decreased by small amount as TCP need time for connection establishment.

10


11/21

Figure: initial throughput Figure: throughput with link 15Mb Figure: throughput with delay 2ms

Figure: throughput with UDP Figure: throughput with 5 nodes

Packet loss:It is one the important parameter as it gives efficiency and reliability of the link.

Various analyses were performed in order to measure packet loss. However with the increased in

link capacity, throughput was increased significantly packet losses were negligible and more

number of packets sent by the link increasing performance of the network. When values of linkdelay were varied, a small amount of change observed in packet losses. As traffic were changed

from TCP to UDP more numbers of packets sent with more packet losses as shown in table

below.

LinkCapacity

LinkDelay

TrafficType

Number ofNodes

PacketSent

PacketLost

1Mb 10ms TCP 3 4324 8

15 Mb 10ms TCP 3 57216 0

11


12/21

1Mb 2ms TCP 3 4390 11

1Mb 10ms UDP 3 11184 822

1Mb 20ms TCP 3 4312 5

1Mb 10ms TCP 5 4370 19

Table: packet losses for different parameters

Delay:

It is mostly used as performance indicator in QoS of real-time application. Link capacity is an

important factor that should be consider during measuring delay. Here when link size was

decreased overall delay were increased to great extend while on increasing link capacity there

was considerable reduction in delay. When link delay decreased to 2ms there was not any major

change in overall delay but on increasing the link delay overall delay was increased considerably.

When traffic is change from UDP to TCP delay decreased considerably and was constant.

Figure: initial delay Figure: delay with link 15Mb Figure: delay with delay 2ms

Figure: delay with UDP Figure: delay with 5 nodes

12


13/21

Jitter:

It is the measure of smoothness of the data transmission, so it should be as low as possible.

Changing link capacity to 15Mb, a significant fall in jitter was observed. Link delay also affected

jitter considerably as it is shown above. As number of nodes increased to 5 nodes, a slight change

was observed in jitter though traffic on link was increased. When traffic type was change to TCPjitter value dropped significantly. The major change observed in TCP was that the jitter remained

constant throughout.

Figure: initial jitter Figure: jitter with link 15Mb Figure: jitter with delay 2ms

Figure: jitter with UDP Figure: jitter with 5 nodes

Exercise 5i:

13


14/21

SFQ and Droptail are two types of queuing policies that we have compared in this exercise. Both

the policies are analyzed for different parameters. Results of which are shown if table below. From

the values from the value it is clear that there is no much difference between them. Only difference

between them is in terms of amount of packet losses where SFQ has more packet loss than

Droptail. When parameters for both queuing policies were changed and performance metric of the

network such as throughput was measured, different results were observed. Graphs for these are

shown below which shows hardly any difference between Droptail and SFQ queuing policies.

Queue type Link capacity Packet size Traffic type Packet lost Packet sent

Drop tail 1MB 500 UDP 403 1402

SFQ 1MB 500 UDP 435 1402

Drop tail .3MB 500 UDP 853 1402

SFQ .3MB 500 UDP 889 1402


SFQ 15MB 500 UDP 0 1402


SFQ 1MB 5000 UDP 412 973

Drop tail 1MB 500 TCP 0 490

SFQ 1MB 500 TCP 0 490

14


15/21

Figure: throughput for droptail queuing policy figure: throughput for SFQ queuing policy

Figure: throughput for droptail with 0.3Mb link figure: throughput for SFQ with 0.3Mb link

Figure: Throughput for Droptail with TCP Figure: Throughput for SFQ with TCP

Exercise 5ii:

In this exercise we are implemented and analyzed RED queue type, then is compared with

previous two types of queuing by measuring different performance metrics of the network. On

analyzing, it is observed that there is no large difference between all three queuing policies. Delay

of RED queue is comparatively smaller and variable. But by comparing it with previous result it

seems that droptail is the best queuing policy amongst all three as it has minimum packet losses.

Queue Type Packet Sent Packet Lost

15


16/21

DropTail 1867 0

SFQ 1867 0

RED 1846 13

Figure: throughput for droptail figure: throughput for SFQ figure: throughput for RED

Figure: Delay for droptail Figure: Delay for SFQ Figure: Delay for RED

Exercise 6:

In this exercise we analyzed routing schemes for network. It is one of the important factor on

which performance of the system depends. Here two routing algorithms viz. Fixed and Adaptive

(DV) were analyzed by using nam simulator. In fixed routing when the link was broken

transmission stops till link gets re-established which is clearly shown in graphs below. In adaptive

routing when link was broken after loss of one packet transmission was routed through another

possible link hence maintaining consistency. Packet losses for both types of routing were one

even though link was broken as traffic used was TCP, Hence on not receiving acknowledgment

after dropped down of link it stopped transmission. It was further analyzed for performance metricsto get more detailed which is shown in graphs below.

16


17/21

Routing type packet transmitted Packet loss

Fixed routing 548 1

Dynamic Routing (DV) 1008 1

Start of simulation Fixed routing after link break Adaptive routing after

link break

Figure: throughput for fixed routing Figure: throughput for adaptive routing

Exercise 8i:

In this exercise comparison between different enhancement techniques for implementing TCPprotocol is carried out. The analysis reveals that performance of the sack is better than Reno and

17


18/21

New Reno technique even though there is not much difference between them. Further verification

is done and all the results are shown in table below and measured performance metrics of the

network. From the graph of throughput it is clear that overall throughput of sack is much constant

than other two.

TCP type Windowsize Delay Throughput Totalnumber ofpackettransmitted

Packet loss

New Reno 24 4 ms 986.4 kbps 1263 16

Reno 24 4 ms 884.8 kbps 1142 15

Sack 24 4 ms 996 kbps 1275 16

New Reno 50 4 ms 980.8 kbps 1266 25

Rano 50 4 ms 881.6 kbps 1152 24

Sack 50 4 ms 992.8 kbps 1282 25

New Reno 24 10 ms 976.8 kbps 1252 12

Reno 24 10 ms 879.2 kbps 1135 11

Sack 24 10 ms 992.8 kbps 1272 13

Figure: throughput for Reno Figure: throughput for NewReno Figure: throughput for Sack

18


19/21


20/21

Shadowing UDP CBR 452 0

Figure: throughput for Two-Ray Ground Figure: throughput for Shadowing

Figure: delay for Two-Ray Ground Figure: delay for Shadowing

Part-III:

Conclusion

Part-I

Part-I of this report gives the overall technology involved in implementing Differential services and

multiprotocol label switching as well as Quality of services for multimedia services in wireless

communication.

20


21/21

References

[1] :[http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=966238&isnumber=20835Giuseppe Bianchi1, Nicola Blefari-

Melazzi2]

[2]: IEEE Communication magazine, may 2002, DiffServ Resource Allocation for Fast Handoff in Wireless Mobile Internet[3]: empirical study of buffer management scheme for Diffserv assured forwarding PHB.

[4] : Meilenstein 3: Preliminary Quasar QoS Architecture,

[5] : E. Rosen, D. Tappan, G. Fedorkow, Y. Rekhter, D. Farinacci, T. Li, A. Conta, MPLS label stack encoding, RFC-3032,January 2001.[6]: Jacquenet C, Bourdon G, Boucadair M, Service Automation and Dynamic Provisioning Techniques in IP/MPLS Environment[7] : E. Rosen, A. Viswanathan, R. Callon, Multiprotocol label switching architecture, RFC 3031, January 2001.[8]: IEEE Communication magazine. December 1999, George Swallow, Cisco Systems

[9] : IEEE Communication magazine. December 1999, MPLS Advantages for Traffic Engineering ,George Swallow, Cisco Systems

[10]: Internet traffic engineering using multi-protocol label switching (MPLS) Daniel O. Awduche a, Bijan Jabbari ][11]: William S, Data and Computer Communication

[12] : IEEE TRANSACTIONS ON MULTIMEDIA, VOL. 3, NO. 4, DECEMBER 2001, Mihaela van der Schaar and Hayder Radha

[13] : J. Lu, Signal processing for internet video streaming: A review, in Proc. IVCP, vol. 2974, Proc. SPIE, Jan. 2000, pp. 246

259

[14]: V. Paxson, End-to-End internet packet dynamics, in Proc. ACM SIG-COM, vol. 27, France, Oct. 1997, pp. 1352

[15] : IEEE magazine, august 2005, CROSS-LAYER DESIGN OF AD HOC NETWORKS FOR REAL- TIME VIDEO STREAMING[16]: A conservative approach to adaptive call admission control for QoS provisioning in multimedia wireless networks , Y.C.Yee

[17]: Supplying Instantaneous Video-on-Demand Services Using Controlled Multicast, Lixin Gao

[18] : Quality of Service for IP Videoconferencing Engineering White Paper, Subha Dhesikan

21

ant final work

Documents